Research map / Learning theory and algorithms / Kernels and regression theory
Kernel Methods in Hyperbolic Spaces
Pengfei Fang, Mehrtash Harandi, Lars Petersson
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| replace Hyperbolic ProtoNet with hyperbolic RBF kernel | ProtoNet (Conv-4) · miniImageNet 5-way 5-shot | accuracy: 72.67 -> 76.09 | Improved |
| replace Hyperbolic ProtoNet with hyperbolic RBF kernel | ProtoNet (Conv-4) · CUB 5-way 5-shot | accuracy: 82.53 -> 85.21 | Improved |
| replace Hyperbolic ProtoNet with hyperbolic RBF kernel | ProtoNet (Conv-4) · tiered-ImageNet 5-way 5-shot | accuracy: 71.96 -> 76.11 | Improved |
| replace Hyperbolic ProtoNet with hyperbolic RBF kernel | ProtoNet (Conv-4) · FC100 5-way 5-shot | accuracy: 51.76 -> 54.40 | Improved |
| hyperbolic tangent kernel instead of Hyperbolic ProtoNet | ProtoNet (ResNet-18) · miniImageNet 5-way 5-shot | accuracy: 76.84 -> 76.65 | No clear effect |
| add hyperbolic tangent kernel to Euclidean baseline | GZSL baseline · GZSL SUN | harmonic mean: 28.5 -> 34.6 | Improved |
| add hyperbolic tangent kernel to Euclidean baseline | GZSL baseline · GZSL CUB | harmonic mean: 26.3 -> 47.9 | Improved |
| add hyperbolic Laplace kernel to Euclidean baseline | GZSL baseline · GZSL AWA2 | harmonic mean: 38.2 -> 66.7 | Improved |
| hyperbolic RBF kernel for knowledge distillation | 4-layer CNN student · CIFAR-100 | accuracy: 49.9 -> 54.4 | Improved |
| hyperbolic RBF kernel instead of Hyperbolic [32] for re-ID | ResNet-50 re-ID · DukeMTMC-reID (dim 32) | mAP: 39.1 -> 46.3 | Improved |
| hyperbolic binomial kernel for re-ID (Market-1501 dim 64) | ResNet-50 re-ID · Market-1501 (dim 64) | R-1: 84.4 -> 83.0 | Hurt |
| use pd hyperbolic kernels instead of indefinite Minkowski kernel (ablation) | hyperbolic few-shot · 5-way 5-shot (4 datasets) | accuracy | Improved |
Closest papers by meaning
- Adaptive Hyperbolic Kernels: Modulated Embedding in de Branges-Rovnyak Spaces (AAAI 2026 · similarity 0.94)
- Random Laplacian Features for Learning with Hyperbolic Space (ICLR 2023 · similarity 0.90)
- Robust large-margin learning in hyperbolic space (NeurIPS 2020 · similarity 0.89)
- Hyperbolic Representation Learning: Revisiting and Advancing (ICML 2023 · similarity 0.89)
- Nested Hyperbolic Spaces for Dimensionality Reduction and Hyperbolic NN Design (CVPR 2022 · similarity 0.89)
- Hyperbolic Embeddings of Supervised Models (NeurIPS 2024 · similarity 0.88)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
